lyupei

hi, thanks for your work for new tensorflow version! I notice that the result of legs on ski image is worse than origin keras version(even with vgg model), do you...

hi I'm interested at the inference time, image resolution on phone, could you share more about this?

Hi, I wanna ask why the outputBufferSize is set as mBindBufferSizes[1] * 6 in line 141 of ctdetNet.cpp? This may cause overflow(in extreme cases) because each detection result contain 15...